skylabsmall lessons. big ideas.Generalattention is all you need
General · A GUIDED PLAYLIST
attention is all you need
Learn attention is all you need through short video lessons.
33 lessons · 25 min · Learn at your pace
Start learningNo account needed. Progress is saved on this device.
Your path through this idea
33 short lessons- 01
Computers can’t read. So how do they translate?
1 min
- 02
Computers don’t read words. They read pieces.
1 min
- 03
Every word becomes a point in space.
1 min
- 04
“Bank” means two things. How does a machine pick one?
1 min
- 05
The old way: read one word at a time.
1 min
- 06
Why the old way forgets your first sentence.
1 min
- 07
Why the old way is painfully slow.
1 min
- 08
What is a GPU? 60 seconds, no jargon.
1 min
- 09
What if every word could look at every other word?
1 min
- 10
Attention is a search engine inside the sentence.
1 min
- 11
How a word finds its match.
1 min
- 12
Turning scores into percentages.
1 min
- 13
How “it” finds “animal.”
1 min
- 14
Where do Q, K, V come from?
1 min
- 15
The tiny √d fix that saves everything.
1 min
- 16
All of attention in one line of math.
1 min
- 17
One look isn’t enough. Use 8.
1 min
- 18
The flaw: it can’t tell “dog bites man” from “man bites dog.”
1 min
- 19
Teaching order with sine waves.
1 min
- 20
The trick that lets you stack many layers.
1 min
- 21
Attention mixes. This part thinks.
1 min
- 22
The encoder: how it reads a sentence.
1 min
- 23
The decoder: writing one word at a time.
1 min
- 24
No peeking: the mask.
1 min
- 25
How the writer looks back at the reader.
1 min
- 26
From numbers back to a word.
1 min
- 27
How does it learn? Guess, check, fix.
1 min
- 28
Start slow, speed up, slow down.
1 min
- 29
Why this trains so much faster.
1 min
- 30
The scoreboard: did it actually work?
1 min
- 31
What do the heads actually look at?
1 min
- 32
One paper. Then ChatGPT.
1 min
- 33
Explain the whole paper in 60 seconds.
1 min